ORGANIC LETTERS
No-D NMR Spectroscopy as a Convenient Method for Titering Organolithium (RLi), RMgX, and LDA Solutions
2004 Vol. 6, No. 15 2567-2570
Thomas R. Hoye,* Brian M. Eklov, and Mikhail Voloshin Department of Chemistry, 207 Pleasant Street, SE, UniVersity of Minnesota, Minneapolis, Minnesota 55455
[email protected] Received May 10, 2004
ABSTRACT
The concentration of organolithium, organomagnesium halide, and lithium diisopropylamide solutions can be reliably determined using No-D NMR spectroscopy by integration against the added internal standard 1,5-cyclooctadiene (COD) (or cyclooctene). In addition, common impurities and degradation products can be assessed.
No-D NMR (or no-deuterium Proton NMR) spectroscopy involves recording 1H NMR spectra of samples dissolved in ordinary, non-deuterium-enriched, laboratory solvents.1 In addition to being useful for in situ monitoring of reaction mixtures, we have found that No-D NMR spectroscopy also provides a convenient and reliable method for determining the concentration of many common reagents. This is especially helpful for air- and moisture-sensitive reagents, like reactive organometallic species. n-Butyllithium is the potential energy source for a plethora of reactions in organic chemistry. Of course, several reliable methods for titration of commercial solutions of alkyllithium and Grignard reagents are known. However, few would argue that this type of quantification is performed as frequently as it should. As we show here, determining the concentration of these solutions by No-D NMR spectroscopy is accurate, easy, and fast. Both Reed and Urwin and Silveira, Jr., Bretherick, Jr., and Negishi developed direct NMR-based (1) Hoye, T. R.; Eklov, B. M.; Ryba, T. D.; Voloshin, M.; Yao, L. J. Org. Lett. 2004, 6, 953-956. 10.1021/ol049145r CCC: $27.50 Published on Web 06/19/2004
© 2004 American Chemical Society
analyses of butyllithiums 25-35 years ago (using added mesitylene or benzene as an internal standard) that appear to have been largely overlooked.2 As the results presented here emphasize, they should not have been. Also, Kamienski has provided an excellent summary of alkyllithium titration methods (including both “wet-” and NMR-based methods).3 In No-D spectroscopy, an aliquot of any solution is simply placed in a conventional NMR sample tube and the 1H NMR data are recorded. Typically, the instrument is in the unlocked mode. Spectra are collected in the same time frame as that of a conventionally locked and shimmed sample. Perfectly adequate levels of signal-to-noise and resolution are routinely attained, in part due to the fact that modern spectrometer hardware exhibits little field drift during data acquisition. (2) (a) Urwin, J. R.; Reed, P. J. J. Organomet. Chem. 1968, 15, 1-5. (b) Reed, P. J.; Urwin, J. R. J. Organomet. Chem. 1972, 39, 1-10. (c) Silveira, A., Jr.; Bretherick, H. D., Jr.; Negishi, E. J. Chem. Educ. 1979, 56, 560-560. (3) Kamienski, C. W. FMC Lithium Link, Winter 1994, Titration Methods for Commercial Organolithium Compounds. http://www.fmclithium.com/ tech/lithiumlinks.asp (accessed May 2004).
Reagent solutions of interest typically have concentrations of 1-2 M. Since most neat organic solvents are ca. 10 M, the solvent:solute ratio is ca. 5-10:1. Not only is it easy to see the solute species of interest, but data can be collected under conditions (see below) that give quantitatively reliable values. The strategy for quantifying NMR samples is simple. A precisely measured amount of a (wisely chosen) standard is combined with a known volume of the solution of interest. The No-D spectrum is recorded and the concentration determined from the integral ratios. We have settled upon the use of 1,5-cyclooctadiene (COD) as an internal integration standard for determination of all of the butyllithiums and the workhorse lithium amide, LDA (lithium diisopropylamide). COD is widely available, stable, dry, easily transferred, and unreactive to RLi/hydrocarbon solutions (at least in the absence of an activator like TMEDA4). It also gives two different proton resonances (vinylic and allylic) for assessing integration self-consistency.5 When the data were collected and analyzed appropriately,6 very reliable ratios (1.00:2.00 ( 0.04) of the two COD resonances were consistently observed. The No-D NMR spectrum of n-BuLi in hexanes containing ca. 12 vol % COD is shown in Figure 1. All of the spectra shown here were recorded from samples prepared in essentially the same manner.7 We determined this solution of (4) (a) Gausing, W.; Wilke, G. Angew. Chem., Int. Ed. Engl. 1978, 17, 371-372. (b) Wetzel, T. G.; Dehnen, S.; Roesky, P. W. Organometallics 1999, 18, 3835-3842. (5) (a) Gerritz, S. W.; Sefler, A. M. J. Comb. Chem. 2000, 2, 39-41. (b) The T1 relaxations for COD (6.5 s for the allylic and 11.2 s for the vinylic resonances) are longer than those measured for the anions (e.g., the n-BuLi C(1)-methylene protons have a T1 relaxation time of 0.85 s). Thus, if the COD integrals are in agreement (i.e., ca. 2.0:1.0), one can safely assume that the anion resonances have fully relaxed as well. (6) Data Acquisition. Spectra were recorded at ambient temperature on Varian (INOVA or VXR) instruments at 300 or 500 MHz. Important parameters for collection of the No-D NMR spectra are the acquisition time, delay, transmitter power, pulse width, and number of transients [for an example where some of these issues were addressed for a sample in a nondeuterated solvent medium (in the context of 1H NMR analysis of a solvolysis rate study) see: Creary, X.; Jiang, Z. J. Org. Chem. 1994, 59, 5106-5108]. Guiding principles are to collect nearly all of the signal (by acquiring magnetization decay for several T1s), to use a sufficiently low transmitter power to avoid baseline artifacts in the transformed spectrum, and to have the total time for the experiment be short. For solutions of relatively high concentration (g0.1 M), signal-to-noise is generally excellent even when a single transient is collected. Once parameters that resulted in relative intensities of COD resonances within 1.00:2.00(0.04 were identified, the protocol was deemed to be sufficient. Values used for all spectra here were at ) 20 s, d1 ) 20 s, tpwr ) 46 and pw ) 7.5 µs (resulting in an ca. 22.5° pulse), and nt ) 2-4. We observed no meaningful differences in the determinations for the same sample using nt ) 1, 2, 4, or 16. Data Workup and Analysis. Care was taken to achieve flat spectral and integral baselines by phasing the spectrum and adjusting the level/tilt of the integral or by performing a baseline correction. Integrals were cut above a flat baseline but inside the 13C satellite peaks (and to avoid the small resonance due to vinylcyclohexene (e0.4%)) contaminant typically present in COD. To assess the impact of subjectivity in interpretation of the integral values, independent analysis of the same COD spectral data set by three different researchers gave allylic to vinylic resonance ratios of 2.011, 2.008, and 1.992 to 1.000 (i.e.,
